28.02.2013 Views

Practical Ship Hydrodynamics

Practical Ship Hydrodynamics

Practical Ship Hydrodynamics

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Introduction 23<br />

experts, while boundary element methods are more widely used. For further<br />

studies, the book by Ferziger and Peric is recommended.<br />

1.5.1 Turbulence models<br />

The RANSE equations require external turbulence models to couple the<br />

Reynolds stresses (terms from the turbulent fluctuations) to the time-averaged<br />

velocities. Turbulence is in general not fully understood. All turbulence<br />

models used for ship flows are semi-empirical. They use some theories about<br />

the physics of turbulence and supply the missing information by empirical<br />

constants. None of the turbulence models used so far for ship flows has been<br />

investigated for its suitability at the free surface. On the other hand, it is<br />

not clear whether an exact turbulence modelling in the whole fluid domain<br />

is necessary for engineering purposes. There are whole books on turbulence<br />

models and we will discuss here only the most primitive turbulence models<br />

which were most popular in the 1990s, especially as they were standard<br />

options in commercial RANSE solvers. ITTC (1990) gives a literature review<br />

of turbulence models as applied to ship flows.<br />

Turbulence models may be either algebraic (0-equation models) or based<br />

on one or more differential equations (1-equation models, 2-equation models<br />

etc.). Algebraic models compute the Reynolds stresses directly by an algebraic<br />

expression. The other models require the parallel solution of additional<br />

differential equations which is more time consuming, but (hopefully) also<br />

more accurate.<br />

The six Reynolds stresses (or more precisely their derivatives) introduce six<br />

further unknowns. Traditionally, the Boussinesq approach has been used in<br />

practice which assumes isotropic turbulence, i.e. the turbulence properties are<br />

independent of the spatial direction. (Detailed measurements of ship models<br />

have shown that this is not true in some critical areas in the aftbody of full<br />

ships. It is unclear how the assumption of isotropic turbulence affects global<br />

properties like the wake in the propeller plane.) The Boussinesq approach then<br />

couples the Reynolds stresses to the gradient of the average velocities by an<br />

eddy viscosity t:<br />

�<br />

u0u0 v0u0 w0u0 � � �<br />

2ux uy C vx uz C wx<br />

u 0 v 0 v 0 v 0 w 0 v 0<br />

u 0 w 0 v 0 w 0 w 0 w 0<br />

D t<br />

⎡<br />

⎣<br />

uy C vx 2vy wy C vz<br />

uz C wx wy C vz 2wz<br />

2<br />

⎤<br />

3 k 0 0<br />

2<br />

0 3 k 0 ⎦<br />

0 0 2<br />

3 k<br />

k is the (average) kinetic energy of the turbulence:<br />

k D 1<br />

2 ⊲u2 C v 2 C w 2 ⊳<br />

The eddy viscosity t has the same dimension as the real viscosity , but<br />

unlike it is not a constant, but a scalar depending on the velocity field. The<br />

eddy viscosity approach transforms the RANSE to:<br />

⊲ut C uux C vuy C wuz⊳ D f1 px<br />

2<br />

3 kx C ⊲ C t⊳⊲uxx C uyy C uzz⊳<br />

C tx2ux C ty⊲uy C vx⊳ C tz⊲uz C wx⊳

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!